Left-ventricular MRI navigation: a novel approach for motion-free thoracic and cardiac radiotherapy

Abstract

Purpose or Objective Cardiac motion adversely affects the accuracy of thoracic and cardiac radiotherapy. Depending on the target location, cardiac motion in excess of 1 cm may pose a larger challenge than respiratory motion. Ideally, radiotherapy treatment delivery is tailored to minimize the effect of both types of motion. A 1D navigator is commonly used to minimize respiratory motion in MR images by measuring the liver/lung interface position to gate each MRI acquisition. Here, we hypothesize that the left-ventricular (LV)/lung interface can be used to capture both cardiac and respiratory motion and that this interface can be used as a 1D navigator structure to obtain motion free end-exhale/diastolic images of the thorax. This would allow cardiac gated imaging without the need for additional cardiac imaging hardware. This is especially appealing for MR-linac treatments, in which integration of the cardiac hardware is not standard and raises concerns about radiation interaction (e.g. hardware malfunctioning). We will quantify the potential benefit of LV navigation over the conventionally used liver/lung navigation with respect to obtaining a motion-free thoracic MRI. Materials and Methods First, real-time (10 Hz) free-breathing 2D coronal TFE images of the heart were acquired on a 1.5T Ingenia (Philips) or a 1.5T Unity MR-linac (Elekta) for 3 healthy volunteers. Next, LV-navigator gated coronal TFE images (2 mm gating window) were acquired of the same plane of the cine scans (Figure 1, top). For benchmarking, a liver/lung navigated gated scans were performed. Residual motion between the dynamics was quantified using rigid registration based on normalized cross correlation on the liver dome and the LV wall. Additionally, the frequency spectrum of the MR-derived navigation-position was computed to verify the presence of both cardiac and respiratory motion components. This was compared to the cardiac and respiratory frequency recorded using MR-compatible physiology monitoring (ECG and respiratory bellow) as independent reference. Results Spectral analysis showed that both respiratory and cardiac frequency components were present in the LV navigator, while only the respiratory motion was sensed by the liver navigator (figure 1). Non-navigated scans had an average motion (20-80% amplitude CC) of the liver dome and LV wall of 8.8 and 6.5 mm, liver navigated scans 1.7 and 7.6 mm and LV navigated scans 1.1 and 2.8 mm, respectively (see figure 2). Conclusion We show that a 1D MRI navigator on the heart can detect simultaneously respiratory and cardiac motion and can be used to reduce both cardiac and respiratory motion during MRI scans. As these navigators are readily available, we envision that this principle can be used to improve pre-beam imaging as well as MR-linac gated treatment strategies in the thorax without the use of external hardware. Using the LV navigator could ensure direct coupling of the image and treatment geometry.